Dashboarding with SAS® and R

Recipe Overview (Steps)

1. Setup your kitchen as explained previously

2. Specify an ODS HTML5 statement to set the file path, name, and format (html in this example)

3. Specify the dashboard title and any font customizations

4. Specify and ODS LAYOUT statement to set the number of rows and columns that will be used to organize the dashboard layout

5. Specify an ODS REGION statement to control where the following output will be placed with respect to the layout (i.e., designating which rows and/or columns each data structure will go to)

6. Include any PROC PRINT, SGPLOT, SGPIE, etc. procedures that encompass the data structures you would like placed in each ODS REGION

7. Specify an ODS LAYOUT END statement to end the dashboard layout

8. Specify an ODS HTML5 CLOSE statement to render the results to the dashboard file (html in this example)

Flavors (Strengths) of SAS

Cooking with SAS^® brings several key flavors (strengths) to the dish:

• ODS syntax is easy to understand and simple to use to define exactly where each data structure should appear on the dashboard.

• PROC SQL is included in base-SAS^®, which offers a simple and flexible query language that can be used to wrangle the data for any table or visual on a dashboard.

• When printing to HTML, the data structures did not become distorted and were easy to update with new underlying data and code.

Aromas (Weaknesses) of SAS

There are also some unique aromas (weaknesses) to consider:

• In this example, PROC SQL queries often included more code than R (tidyverse) equivalents.

• SAS^® Studio via SAS^® OnDemand for Academics (ODA) includes a way of saving data through temporary cloud libraries, which can be hard to manage with multiple datasets and when internet disconnections occur.

• Some commonly used visualizations, such as a word cloud, are not available in base-SAS^® and thus require more investment in SAS^® software to utilize.

Setting up Your R Kitchen

• This demonstration utilized R v.4.2.3 and was accessed on RStudio^® as the primary Integrated Development Environment (IDE).

• As with any project in RStudio^®, a primary folder should be setup that will be used as the working directory. This is where the R project (Rproj) file will be housed, as well as any required subfolders.

• Within RStudio^®, only four libraries are installed and imported -readxl, which is used to read in excel spreadsheet files, quarto, which is the publishing system used to generate the dashboard, tidyverse, which is used for data wrangling and visualization, and gt, which is used for the generation of tables.

• The two excel spreadsheets (Publications and Authors) were imported as dataframes (PUBS and AUTHORS).   

Recipe Overview (Steps)

1. Setup your kitchen as explained previously

2. In RStudio^®, create a new Quarto^® Document with the minimum YAML header to ensure proper HTML setup for the dashboard:

---

format: html

embed-resources: true

page-layout: custom

editor: visual

colorlinks: TRUE

---

3. Create an R code block for setup and data wrangling. This block is used for loading the libraries, reading in the data (using readxl), and wrangling in preparation for the generation of tables and plots (using tidyverse). The output of this block should not be included in the dashboard output.
4. Create an additional code block that will serve as the layout for the dashboard. This is accomplished using the following code cell options (adjusting numbers to fit the needs of the dashboard):

#| layout-ncol: 3

#| layout-nrow: 3

5. Within this code block, tables can be generated using gt and plots using ggplot2 (or any other table/plot package in R). As data structures are generated, they will occupy the spaces within 3x3 layout.
6. Render the document within RStudio^® to write the HTML document, which will be saved to the working directory and opened automatically (typically within a web browser).

Aromas (Weaknesses) of R

However, there are also some unique aromas (weaknesses) to consider:

• Quarto^® syntax is easy to understand, but it is harder to specify the exact location of data structures on the dashboard compared to SAS^® ODS, which created extra white space.

• When printing to HTML, the data structures did become distorted, and many trials were needed to get the correct resolution to output.

• Some visualizations, such as pie charts, are much more complicated to generate compared to something like SGPIE within SAS^®.

• FREQUENT updates; pre-release 1.4

Conclusions

• Dashboards are key analytic tools that are used by businesses, schools, and throughout the healthcare industry.

• These tools are defined by their ability to utilize frequently regenerated and often real-time data to inform decision makers.

• There are many programming languages that can be used to create a dashboard, akin to the use of different ingredients used during cooking. This paper explored two popular ingredients (i.e., programming languages) that are used to generate dashboards, SAS^® and R.

• This paper was a basic introduction to dashboarding with these two languages. More advanced dashboards are possible (ex: Quarto pre-release, Shiny).